Multi-stage coding block partition search is disclosed. A method includes selecting a partition-none partition type and a partition-split partition type for predicting the block, determining a first cost of predicting the block using the partition-none partition type, and determining a second cost of predicting the block using the partition-split partition type. The partition-none partition type and the partition-split partition type are selected from a set of partition types that includes the partition-none partition type, the partition-split partition type, and third partition types. The method also includes, on condition that the result meets a criterion, determining a respective encoding cost corresponding to at least some of the third partition types; selecting a selected partition type corresponding to a minimal cost amongst the partition-none partition type and the at least some of the third partition types; and encoding, in a compressed bitstream, the selected partition type.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A method for predicting a block of size N×N of a video frame, comprising: selecting a partition-none partition type and a partition-split partition type for predicting the block, wherein the partition-none partition type and the partition-split partition type are selected from a set of partition types comprising the partition-none partition type, the partition-split partition type, and third partition types, the partition types being a same level partitions, wherein the partition-none partition type includes one prediction unit of size N×N corresponding to the block of size N×N, and wherein the partition-split partition type partitions the block into equally sized square sub-blocks, each of square sub-blocks having a size of N/2×N/2; determining a first cost of predicting the block using the partition-none partition type; determining a second cost of predicting the block using the partition-split partition type; determining a result of comparing the first cost and the second cost; on condition that the result meets a criterion indicating that the partition-none partition type is preferred over the partition-split partition type: determining a respective encoding cost corresponding to at least some of the third partition types; and selecting a selected partition type corresponding to a minimal cost amongst the partition-none partition type and the at least some of the third partition types; and encoding, in a compressed bitstream, the selected partition type.
Video compression technology for predicting a block of a video frame. The problem addressed is efficiently selecting the best partitioning strategy for a block to minimize encoding cost. The method involves predicting a block of size N×N. Two primary partition types are considered: a "partition-none" type, which uses a single prediction unit of size N×N for the entire block, and a "partition-split" type, which divides the block into four equally sized square sub-blocks, each of size N/2×N/2. These are selected from a set of partition types at the same hierarchical level. A cost is determined for predicting the block using the partition-none type, and another cost is determined for the partition-split type. These costs are compared. If the comparison indicates that the partition-none type is preferred over the partition-split type, then encoding costs are determined for at least some additional "third partition types." These third partition types are also at the same hierarchical level as partition-none and partition-split. Finally, a selected partition type is chosen that corresponds to the minimal cost among the partition-none type and the evaluated third partition types. The selected partition type is then encoded into a compressed bitstream.
2. The method of claim 1 , further comprising: on condition that the result does not meet the criterion: selecting the partition-split partition type as the selected partition type.
A method for optimizing data partitioning in a database system addresses the challenge of efficiently organizing data to improve query performance. The method involves evaluating different partition types to determine the most effective way to split data across storage structures. Initially, the method assesses a dataset to identify an optimal partition type based on predefined criteria, such as query performance, storage efficiency, or load balancing. If the initial partition type does not meet the criteria, the method automatically selects a partition-split partition type as the alternative. The partition-split type involves dividing data into smaller, more manageable segments while maintaining logical relationships between the segments. This approach ensures that data is distributed in a way that minimizes access time and maximizes resource utilization. The method dynamically adjusts partitioning strategies based on real-time performance metrics, allowing the database system to adapt to changing workloads and data characteristics. By incorporating this adaptive partitioning technique, the system enhances overall efficiency and scalability, particularly in large-scale database environments where traditional partitioning methods may fall short.
3. The method of claim 1 , wherein determining the second cost of predicting the block using the partition-split partition type comprises: recursively partitioning, based on a quad-tree partitioning, the block into sub-blocks using the partition-split partition type.
This invention relates to video encoding and decoding, specifically improving the efficiency of block partitioning in predictive coding. The problem addressed is the computational cost and inefficiency in determining optimal partition types for video blocks during encoding, particularly when using quad-tree partitioning. Traditional methods often require excessive processing to evaluate different partition types, leading to delays and suboptimal encoding decisions. The invention provides a method for determining the cost of predicting a video block using a partition-split partition type, which involves recursively partitioning the block into sub-blocks based on quad-tree partitioning. The quad-tree partitioning divides the block into smaller sub-blocks in a hierarchical manner, where each sub-block can be further partitioned if needed. The method evaluates the cost of predicting the block by assessing the partitioning process, which helps in selecting the most efficient partition type for encoding. This approach reduces redundant computations by leveraging recursive partitioning, ensuring that only necessary sub-blocks are processed, thereby improving encoding efficiency and performance. The invention is particularly useful in video compression standards that rely on quad-tree partitioning, such as HEVC and VVC, where optimal partitioning decisions are critical for achieving high compression efficiency.
4. The method of claim 1 , wherein the third partition types comprise a partition_vert partition type and a partition_horz partition type.
This invention relates to data partitioning in a database system, specifically for optimizing query performance by dividing data into partitions based on spatial or geometric criteria. The problem addressed is inefficient data retrieval in large datasets where traditional partitioning methods do not account for spatial relationships, leading to slow query performance and excessive resource usage. The method involves partitioning data into multiple partitions, where each partition is assigned a specific partition type. The partition types include a partition_vert type, which divides data along vertical boundaries, and a partition_horz type, which divides data along horizontal boundaries. These partition types allow for efficient spatial indexing and querying, enabling faster retrieval of data based on geometric or spatial criteria. The partitioning is dynamically adjusted based on query patterns and data distribution to maintain optimal performance. The method also includes determining the optimal partition boundaries and assigning data to the appropriate partitions based on their spatial attributes. This ensures that queries involving spatial operations, such as range queries or nearest-neighbor searches, can be executed efficiently by accessing only the relevant partitions. The system dynamically updates the partition types and boundaries as new data is added or as query patterns change, ensuring continuous performance optimization. This approach improves query speed and reduces computational overhead in database systems handling large-scale spatial data.
5. The method of claim 4 , wherein the third partition types further comprise a partition_horz_a partition type, a partition_horz_b partition type, a partition_vert_a partition type, a partition_vert_b partition type, a partition_horz_4 partition type, and a partition_vert_4 partition type.
This invention relates to video encoding and decoding, specifically partitioning techniques for block-based video compression. The problem addressed is improving compression efficiency by providing more flexible partitioning options for video blocks, allowing better adaptation to varying content characteristics. The method involves defining multiple partition types for dividing a video block into smaller sub-blocks. These partitions include horizontal and vertical splits, as well as more complex configurations. Specifically, the partition types include horizontal splits (partition_horz_a and partition_horz_b), vertical splits (partition_vert_a and partition_vert_b), and uniform four-part splits (partition_horz_4 and partition_vert_4). The partition_horz_a and partition_vert_a types divide the block into two equal sub-blocks, while partition_horz_b and partition_vert_b allow asymmetric splits. The partition_horz_4 and partition_vert_4 types divide the block into four equal sub-blocks, either horizontally or vertically. These partitioning options enable more precise adaptation to local content features, improving compression efficiency by reducing redundancy while maintaining visual quality. The method is particularly useful in advanced video coding standards where flexible partitioning is critical for handling diverse video content.
6. The method of claim 1 , wherein the block has a size of 128×128.
This invention relates to image processing, specifically a method for handling image data in blocks of a fixed size. The problem addressed is the need for efficient and standardized image data processing, particularly in applications like compression, encoding, or analysis, where consistent block sizes improve performance and compatibility. The method involves dividing an image into blocks, each with a predefined size of 128×128 pixels. This block size is chosen to balance computational efficiency and data granularity, ensuring that processing tasks such as transformation, filtering, or encoding can be performed uniformly across the image. The fixed block size simplifies hardware and software implementation, as it allows for parallel processing and optimized memory access patterns. The method may be part of a larger image processing pipeline, where the 128×128 blocks are further processed using techniques such as discrete cosine transforms, wavelet transforms, or machine learning-based analysis. The consistent block size ensures that these subsequent steps can be applied predictably, reducing complexity and improving throughput. This approach is particularly useful in applications like video encoding, medical imaging, or satellite imagery, where large datasets must be processed efficiently while maintaining high accuracy. The fixed block size also facilitates standardization, making it easier to integrate with existing systems and protocols.
7. The method of claim 1 , wherein the block has a size of 64×64.
The invention relates to image processing, specifically to a method for encoding or decoding image data using block-based compression. The problem addressed is improving compression efficiency and processing speed by optimizing block size during image encoding or decoding. The method involves dividing an image into blocks for processing, where each block has a predefined size. In this specific implementation, the block size is fixed at 64×64 pixels. The method may include steps such as transforming the block data, quantizing the transformed data, and entropy encoding the quantized data to achieve compression. The use of a 64×64 block size balances computational efficiency with compression performance, particularly for high-resolution images or video frames. The method may also include inverse operations for decoding, such as inverse transforming, inverse quantizing, and entropy decoding the compressed data to reconstruct the original image. The fixed block size simplifies hardware implementation and reduces computational overhead compared to variable block sizes. This approach is particularly useful in applications requiring real-time processing, such as video streaming or high-definition image transmission.
8. The method of claim 1 , wherein the criterion comprises the first cost being less than the second cost.
A method for optimizing resource allocation in a distributed computing system addresses the problem of inefficient resource utilization due to suboptimal cost-based decision-making. The method evaluates multiple resource allocation options by comparing their associated costs and selects the option that meets a predefined cost criterion. Specifically, the method determines a first cost for a first resource allocation option and a second cost for a second resource allocation option. The criterion for selection is that the first cost must be less than the second cost, ensuring the most cost-effective option is chosen. This approach helps reduce operational expenses and improves system efficiency by dynamically selecting the lowest-cost resource allocation strategy. The method can be applied in cloud computing environments, data centers, or any system where resource allocation decisions impact cost. By systematically comparing costs and enforcing a strict cost-based selection rule, the method ensures optimal resource utilization while minimizing financial overhead.
9. The method of claim 1 , wherein the criterion comprises the first cost being within a predefined range of the second cost.
This invention relates to cost comparison systems for optimizing financial decisions. The problem addressed is the difficulty in evaluating and comparing costs across different options to ensure cost-effective choices. The invention provides a method for comparing costs by establishing a criterion that ensures the first cost is within a predefined range of the second cost, enabling users to make informed decisions based on cost proximity. The method involves analyzing two or more cost values and determining whether the difference between them falls within an acceptable range. This predefined range can be set based on user preferences, industry standards, or other relevant factors. By ensuring the first cost is within this range of the second cost, the system helps users avoid significant cost discrepancies, promoting financial efficiency. The method may also include additional steps such as retrieving cost data from multiple sources, normalizing the data for comparison, and applying the predefined range to assess cost alignment. The system can be used in various applications, including financial planning, procurement, and budgeting, where cost consistency is critical. The invention improves decision-making by providing a structured approach to cost evaluation, reducing the risk of overspending or selecting suboptimal options.
10. An apparatus for predicting a block of size N×N of a video frame, comprising: a memory; and a processor, the processor configured to execute instructions stored in the memory to: select a partition-none partition type and a partition-split partition type for predicting the block, wherein the partition-none partition type and the partition-split partition type are selected from a set of partition types comprising the partition-none partition type, the partition-split partition type, and third partition types, wherein the partition-none partition type includes one prediction unit of size N×N corresponding to the block of size N×N, and wherein the partition-split partition type partitions the block into equally sized square sub-blocks, each of square sub-blocks having a size of N/2×N/2; determine a first cost of predicting the block using the partition-none partition type; determine a second cost of predicting the block using the partition-split partition type; determine a result of comparing the first cost and the second cost; on condition that the result meets a criterion indicating that the partition-none partition type is preferred over the partition-split partition type: determine a respective encoding cost corresponding to at least some of the third partition types; and select a selected partition type corresponding to a minimal cost amongst the partition-none partition type and the at least some of the third partition types.
The invention relates to video frame prediction in video encoding, specifically for selecting an optimal partition type for predicting an N×N block of a video frame. The problem addressed is efficiently determining the best partition strategy to minimize encoding cost while maintaining prediction accuracy. The apparatus includes a memory and a processor that executes instructions to evaluate different partition types for the block. The partition types include a partition-none type, which treats the entire block as a single prediction unit of size N×N, and a partition-split type, which divides the block into four equally sized square sub-blocks of size N/2×N/2. The processor calculates the prediction cost for both the partition-none and partition-split types. If the partition-none type is determined to be preferable based on a cost comparison, the processor further evaluates the encoding costs of additional partition types (third partition types) and selects the partition type with the minimal cost among the partition-none type and the evaluated third partition types. This approach optimizes the prediction process by dynamically selecting the most efficient partition strategy based on cost analysis.
11. The apparatus of claim 10 , wherein the instructions further comprise: on condition that the result does not meet the criterion: selecting the partition-split partition type as the selected partition type.
A system for optimizing data partitioning in a database or data processing environment addresses the challenge of efficiently organizing data to improve query performance and storage efficiency. The system includes a processor and memory storing instructions that, when executed, perform operations to evaluate and select an optimal partition type for a dataset. The instructions analyze the dataset to determine whether it meets a predefined criterion, such as a threshold for data distribution, query performance, or storage efficiency. If the criterion is not met, the system automatically selects a partition-split partition type, which involves dividing the dataset into smaller, more manageable partitions based on specific criteria like key ranges or hash values. This selection process ensures that the data is partitioned in a way that enhances query speed, reduces storage overhead, and maintains data integrity. The system may also include additional logic to dynamically adjust partitioning strategies based on changing data characteristics or workload patterns, ensuring continuous optimization. By automating the partition selection process, the system reduces manual intervention and improves overall database performance.
12. The apparatus of claim 10 , wherein determining the second cost of predicting the block using the partition-split partition type comprises: recursively partitioning, based on a quad-tree partitioning, the block into sub-blocks using the partition-split partition type.
This invention relates to video encoding and decoding, specifically improving efficiency in block partitioning for predictive coding. The problem addressed is the computational overhead and suboptimal prediction accuracy when determining the cost of using partition-split partitioning (a method where a block is divided into smaller sub-blocks) in video compression. The invention provides a method to evaluate the cost of predicting a block using partition-split partitioning by recursively dividing the block into sub-blocks using quad-tree partitioning. Quad-tree partitioning splits a block into four smaller sub-blocks, and this process is repeated recursively to further subdivide the sub-blocks. The cost of predicting the block is then determined based on the resulting sub-blocks. This approach improves prediction accuracy by allowing finer granularity in partitioning while reducing computational complexity by leveraging hierarchical quad-tree structures. The invention is particularly useful in video codecs where efficient block partitioning is critical for achieving high compression ratios and maintaining visual quality.
13. The apparatus of claim 10 , wherein the third partition types comprise a partition_vert partition type and a partition_horz partition type.
This invention relates to a data processing apparatus for managing memory partitions in a computing system. The apparatus includes a memory controller configured to allocate and manage memory partitions of different types to optimize data storage and access efficiency. The partitions are categorized into at least three types: a partition_vert type, a partition_horz type, and a third partition type. The partition_vert type organizes data in a vertical arrangement, while the partition_horz type arranges data horizontally. The third partition type may include additional configurations to support specific data access patterns or performance requirements. The memory controller dynamically assigns these partition types based on workload characteristics, ensuring efficient memory utilization and reducing access latency. The apparatus further includes mechanisms to monitor partition performance and reallocate partitions as needed to maintain optimal system efficiency. This invention addresses the challenge of efficiently managing memory resources in computing systems with varying data access patterns, improving overall system performance and energy efficiency.
14. The apparatus of claim 10 , wherein the third partition types further comprise a partition_horz_a partition type, a partition_horz_b partition type, a partition_vert_a partition type, a partition_vert_b partition type, a partition_horz_4 partition type, and a partition_vert_4 partition type.
This invention relates to video encoding and decoding, specifically to partitioning schemes for block-based video compression. The problem addressed is the need for efficient and flexible partitioning of video blocks to improve compression efficiency while maintaining computational feasibility. The invention introduces additional partition types to enhance the granularity and adaptability of block partitioning during encoding and decoding processes. The apparatus includes a video encoder or decoder configured to process video blocks using multiple partition types. These partition types include horizontal and vertical splits, as well as more complex divisions such as horizontal and vertical quadtree splits. The additional partition types—partition_horz_a, partition_horz_b, partition_vert_a, partition_vert_b, partition_horz_4, and partition_vert_4—provide further flexibility in dividing blocks into smaller sub-blocks. These partitions allow for finer control over block shapes, enabling better adaptation to local video content characteristics. The encoder selects the optimal partition type based on rate-distortion optimization, while the decoder reconstructs the video using the same partition types to ensure consistency. This approach improves compression efficiency by reducing redundancy and better matching the partitioning to the video's spatial and temporal characteristics.
15. The apparatus of claim 10 , wherein the block has a size of 128×128.
The invention relates to a data processing apparatus for handling image or video data, specifically focusing on block-based processing techniques. The apparatus includes a processing unit configured to divide input data into blocks of a predefined size for further analysis or transformation. A key aspect of the invention is the use of a block size of 128×128 pixels, which optimizes computational efficiency and memory access patterns during processing. The apparatus may also include a memory module to store intermediate results and a control unit to manage data flow between components. The block size selection is particularly advantageous for applications requiring high-throughput processing, such as real-time video encoding or image compression, where minimizing latency and maximizing parallelism are critical. The apparatus may further incorporate error correction or data validation mechanisms to ensure integrity during block-based operations. The invention addresses the challenge of balancing computational load and memory bandwidth in block-based processing systems, providing a standardized block size that enhances performance in various multimedia applications.
16. The apparatus of claim 10 , wherein the block has a size of 64×64.
This invention relates to image processing, specifically to systems for encoding and decoding image data using block-based compression techniques. The problem addressed is improving compression efficiency while maintaining image quality, particularly in applications like video streaming or storage where bandwidth and storage constraints are critical. The apparatus includes a block-based image processor that divides an image into smaller blocks for compression. A key feature is the use of a block size of 64×64 pixels, which balances computational efficiency and compression performance. The processor applies a transform, such as a discrete cosine transform (DCT), to convert spatial domain data into frequency domain coefficients. These coefficients are then quantized and entropy-encoded to reduce redundancy. The apparatus may also include a motion compensation module for video processing, which predicts motion between frames using the 64×64 blocks. This reduces temporal redundancy by encoding only the differences between frames. The system further includes a reconstruction module that reverses the encoding process for decoding, ensuring accurate image reconstruction. The 64×64 block size is chosen to optimize transform efficiency, as larger blocks capture more spatial correlations but require more computation. This size also aligns with common video coding standards, ensuring compatibility. The apparatus may be implemented in hardware, software, or a combination, suitable for real-time applications. The invention improves compression ratios while maintaining visual quality, making it useful for high-definition video streaming and storage.
17. The apparatus of claim 10 , wherein the criterion comprises the first cost being less than the second cost.
A system for optimizing resource allocation in a distributed computing environment addresses the challenge of efficiently managing computational tasks across multiple nodes to minimize costs while maintaining performance. The system evaluates the costs associated with executing a task on different nodes and selects the node with the lowest cost that meets predefined performance criteria. The cost comparison involves assessing factors such as energy consumption, processing time, and financial expenses. The apparatus includes a cost estimation module that calculates the first cost for executing a task on a primary node and the second cost for executing the same task on an alternative node. The decision-making module then compares these costs and selects the node where the first cost is less than the second cost, ensuring cost-effective task execution. This approach optimizes resource utilization by dynamically allocating tasks to the most economical node without compromising performance, making it particularly useful in cloud computing and large-scale distributed systems. The system enhances efficiency by reducing unnecessary computational expenses while maintaining system reliability and responsiveness.
18. The apparatus of claim 10 , wherein the criterion comprises the first cost being within a predefined range of the second cost.
A system for optimizing resource allocation in a distributed computing environment addresses the challenge of efficiently distributing tasks across multiple computing nodes while minimizing operational costs. The system evaluates the costs associated with different resource allocation strategies and selects an optimal configuration based on predefined criteria. Specifically, the system compares the cost of using a first set of computing resources with the cost of using a second set of resources. The selection criterion includes ensuring that the first cost falls within a predefined range of the second cost, thereby balancing cost efficiency with performance requirements. The system dynamically adjusts resource allocation in real-time to adapt to changing workload demands and cost constraints. This approach reduces unnecessary resource usage while maintaining service quality, making it particularly useful in cloud computing and large-scale data processing environments. The system may also incorporate additional factors such as latency, reliability, and energy consumption to further refine the allocation decisions. By automating the cost-based selection process, the system ensures that resources are allocated in a manner that aligns with budgetary limits while meeting performance objectives.
19. An apparatus for predicting a block of size N×N of a video frame, comprising: a memory; and a processor, the processor configured to execute instructions stored in the memory, the instructions comprising: determining a partition type, from partition types comprising a partition-none partition type, a partition-split partition type, and third partition types, for predicting the block by operations comprising: determining a first coding cost of the block associated with the partition-none partition type, wherein the partition-none partition type includes one prediction unit of size N×N corresponding to the block of size N×N; determining a second coding cost of the block associated with a skip-level recursive partitioning, wherein the skip-level recursive partitioning partitions the block into square sub-blocks, and wherein each sub-block having a size that is less than N/2×N/2; on condition that the first coding cost is smaller than the second coding cost indicating that the partition-none partition type is preferred over the skip-level recursive partitioning: determining respective coding costs of encoding the block using at least some of the third partition types and the partition-split partition type; and selecting the partition type corresponding to a minimal coding cost from among the first coding cost and the respective coding costs; and encoding, in a compressed bitstream, the partition type.
This invention relates to video encoding, specifically to predicting blocks of size N×N in a video frame. The problem addressed is efficiently selecting an optimal partition type for encoding a block to minimize coding cost while maintaining video quality. The apparatus includes a memory and a processor that executes instructions to determine the best partition type for predicting the block. The partition types include a partition-none type, which uses a single prediction unit of size N×N, and a partition-split type, which recursively divides the block into smaller square sub-blocks. The processor first compares the coding cost of the partition-none type with a skip-level recursive partitioning that splits the block into sub-blocks smaller than N/2×N/2. If the partition-none type has a lower cost, the processor evaluates additional partition types, including the partition-split type, to determine which yields the minimal coding cost. The selected partition type is then encoded into a compressed bitstream. This approach optimizes encoding efficiency by dynamically selecting the most cost-effective partition strategy for each block.
20. The apparatus of claim 19 , wherein the third partition types comprise a partition_vert partition type and a partition_horz partition type.
This invention relates to a data processing apparatus for managing memory partitions in a computing system. The apparatus includes a memory controller configured to allocate and manage memory partitions of different types to optimize data storage and retrieval efficiency. The partitions are categorized into at least three types: a partition_vert type, a partition_horz type, and a third partition type. The partition_vert type organizes data in a vertical arrangement, while the partition_horz type arranges data horizontally. The third partition type may include additional configurations to further enhance memory management. The memory controller dynamically assigns these partition types based on data access patterns, reducing latency and improving system performance. The apparatus also includes a monitoring module to track memory usage and adjust partition allocations in real-time. This system addresses the challenge of inefficient memory utilization in computing systems by providing flexible and adaptive partitioning strategies. The invention aims to optimize memory access speeds and reduce fragmentation, particularly in systems handling large datasets or complex workloads.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
June 25, 2018
January 7, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.